Inhibition of fibroblast activation protein partially restores the homeostatic phenotype of mechanoactivated synoviocytes

Abstract

Objective:

Synovial fibrosis, characterized by mechanoactivation of fibroblast-like synoviocytes (FLS), is thought to be a driver of osteoarthritis (OA). Fibroblast activation protein (FAP) is expressed by activated FLS and contributes to OA progression, but the mechanisms by which it does so are undefined. The objective of this study was to evaluate FAP expression in OA across species, assess FAP regulation in FLS, and determine how FAP inhibition alters the phenotype of mechanoactivated FLS.

Methods:

Synovium from human patients (n = 9) and animal models of OA (murine n = 5; porcine n = 8) was evaluated for FAP expression via immunofluorescence. An in vitro model of FLS mechanoactivation was established by plating FLS on polyacrylamide gels or glass +/−TGFβ (n = 5/group). Activated FLS were treated with a FAP inhibitor (FAPi) and phenotype was evaluated via immunofluorescence, RT-qPCR, collagen gel contraction, and FUNCAT (n = 5–7/group).

Results:

FAP was increased in mouse (MD: 10.3, 95% CI: [3.0, 17.6]), pig (8.7 [4.6, 12.8]), and human (24.5 [11.8, 37.3]) OA synovium and correlated with histological markers of synovial pathology. In vitro culture on stiff substrates activated FLS, and treatment with TGFβ induced FAP expression (1.8 [0.3, 3.4]). FAP inhibition partially rescued the morphological, transcriptional, and functional profile of activated FLS (e.g., reduced stress fiber formation (−0.2 [−0.35, −0.05]), αSMA expression (−2.9 [−4.9, −0.84], and contractility (−0.16 [−0.33, −0.0]).

Conclusions:

The upregulation of FAP in OA synovium is conserved across species. FAP is a TGFβ-responsive driver of FLS mechanoactivation, and its inhibition partially restores the homeostatic function of FLS. Targeting FAP may represent a novel strategy to improve synovial health and mitigate joint degeneration in OA.

Introduction

Osteoarthritis (OA) is a painful and debilitating chronic degenerative joint disease [1]. The synovium, a bilayer membrane that lines joints and produces synovial fluid [2], undergoes substantial structural and functional changes during OA progression [3], making it an appealing target for intervention. Under healthy conditions, the thin inner layer (intima) of the synovium is populated by fibroblast-like synoviocytes (FLS) and macrophages, while the thicker outer layer (subintima) is composed of a loose vascularized connective tissue with few cells [4]. Synovitis and synovial damage are well-recognized features of OA [5], and radio-graphic studies show that synovial pathology may predict future development of cartilage lesions, suggesting that the synovium could play a role in the development of OA [6]. While the initial trigger of synovial inflammation remains unclear, many have hypothesized that cartilage matrix fragments released due to injury or chronic degeneration activate synovial macrophages and FLS. Synovial cells then release inflammatory cytokines, activating chondrocytes and upregulating matrix-metallo-proteinases (MMPs), leading to further cartilage degeneration and initiating a vicious cycle of inflammation and catabolism [7].

Recent work from our lab and others shows that the stiffness of synovial tissue increases during OA progression [8–10]. As the synovial extracellular matrix stiffens and inflammatory signaling increases, FLS become “activated” and differentiate towards myofibroblasts: contractile cells characterized by mature actin stress-fibers and alpha smooth muscle actin (αSMA) expression [11–13]. Myofibroblasts produce inflammatory cytokines, including IL-1β, IL-6, and TNF-α, as well as copious extracellular matrix (ECM), which is compacted by the increasingly contractile FLS, further contributing to fibrosis and synovial stiffening [11,14]. The critical role of activated FLS in propagating fibrosis and inflammation makes them an ideal therapeutic target, and our prior work showed that suppression of FLS contractility (using the Rho/ROCK inhibitor Fasudil) restores production of lubricin, one of the key homeostatic functions of the synovium [15].

In addition to Rho/ROCK mediated contractility, there are several other molecular pathways and proteins present in activated synovium that may play a role in disease progression. One such molecule, fibroblast activation protein (FAP), is a serine protease expressed at low levels in homeostatic conditions, but upregulated in states of inflammation and fibrosis [16]. Using a global knockout mouse (gKO), Fan et al. [17] showed that FAP may directly contribute to OA progression after destabilization of the medial meniscus (DMM) by degrading collagen I and collagen II after initial cleavage by MMPs [17]. While the majority of FAP is membrane-associated, a portion of the protein can be shed from the cell surface in a soluble form [18]. In their study, Fan et al. [17] attributed the improvement in joint health seen in FAP gKO mice to decreased collagen degradation by FAP secreted into the synovial fluid, and further demonstrated that small molecule inhibition via intraarticular injection of a FAP inhibitor (Ac-Gly-BoroPro, FAPi) could slow disease progression.

While FAP can clearly play a role in cartilage breakdown via its proteolytic activity, membrane-associated FAP may contribute to additional cellular behaviors, including mechanoactivation. Specifically, membrane-associated FAP promotes migration, proliferation, and invasion of multiple cancer cell types, as well as bone marrow mesenchymal stromal cells [19–21], and may do so independent of its catalytic function. Studies have also reported colocalization of FAP with integrin α₃β₁ and involvement of FAP in collagen-binding [22–24]. These findings support a potential role for FAP in regulating mechanobiological pathways and sensation of the microenvironment, and suggest that FAP may also contribute to OA pathogenesis by promoting synovial cell mechanoactivation. The purpose of this study was to confirm the upregulation of FAP in OA across species, evaluate how FAP expression is regulated during FLS activation, and determine if and how FAP inhibition alters the phenotype of mechanoactivated FLS. We hypothesized FAP would be upregulated in mechanoactivated FLS and that inhibition of FAP would rescue the homeostatic phenotype of these cells – here defined by the production of lubricin and the absence of myofibroblastic and fibrotic features (e.g. low αSMA expression, reduced contractility, and decreased profibrotic gene expression).

Methods

Synovial immunofluorescence and histology

See Supplemental methods for information on sample collection (human, porcine, and murine), immunofluorescent staining, histology, and histological scoring.

Fibroblast-like synoviocyte isolation and culture

Primary fibroblast-like synoviocytes (FLS) were isolated from the suprapatellar pouch of juvenile bovine stifle joints (n = 7) as previously described [15]. Briefly, synovium was digested in 0.1% Type IV collagenase (Worthington Biochemical, LS004189), filtered through a 70μm cell strainer, and plated on tissue culture-treated polystyrene in basal media (DMEM with 10% fetal bovine serum and 1% penicillin/streptomycin/fungizone). Cells were used for experiments at passages 1–3 (P1-P3).

To establish an in vitro model of FLS mechanoactivation, which we define as a myofibroblast-like state characterized by increased contractility, cytoskeletal remodeling, focal adhesion maturation, and profibrotic gene expression, cells were seeded onto 15kPa polyacrylamide (PA) gels (soft substrate) or glass slides (stiff substrate) (n = 10/group). Half the samples on each substrate were additionally treated with TGFβ−1 (10ng/mL, Invitrogen PHG9204) (n = 5/group) to amplify the myofibroblastic phenotype [25]. PA gels were fabricated according to our published procedures [15], functionalized with sulfo-SANPAH (ThermoFisher Scientific 22589), and conjugated with fibronectin (20μg/mL). Cells were seeded at 50,000 cells/well on PA gels and 30,000 cells/well on glass slides in 4-well plates. After 48 h, cells were harvested for RT-qPCR or fixed and permeabilized in microtubule stabilizing buffer (2% paraformalde-hyde, 0.1 M PIPES (pH 6.75), 1 mM EGTA, 1 mM MgSO₄, 4% (w/v) poly (ethylene glycol), 1% Triton X-100) prior to staining.

For experiments with FAPi and Fasudil, FLS were mechanoactivated by plating on glass (30,000 cells/well) with or without TGFβ−1 (10ng/mL). FAPi (n = 5, Ac-Gly-BoroPro, 10μg/mL, MedChemExpress HY-101801), Fasudil (n = 4, 10μM, Thomas Scientific CDS021620), and TGFβ−1 were added on day 0 and day 3. On day 6, cells were harvested for RT-qPCR and fixed for staining as previously described.

In vitro immunofluorescence and image analysis

See Supplemental methods for information on in vitro immunofluorescence and analysis of cell morphology.

Functional assays

See Supplemental methods for information on collagen gel contraction assays and nascent matrix labelling.

Statistical analyses

FAP and αSMA staining intensity data for murine, porcine, and human synovium were tested for normality using Shapiro-Wilk tests. Normally distributed, paired data were compared via paired t-test. Normally distributed, unpaired data were compared via Welch’s t test. Non-normally distributed data were compared by Mann-Whitney test. Differences in FAP and αSMA staining intensity in porcine synovium at 6 weeks and 6 months were evaluated using linear mixed-effects models with fixed effects for injury status, time point, and their interaction, and a random intercept for animal. Post-hoc pairwise comparisons were performed using t-tests of estimated marginal means with Tukey adjustment. Residuals were evaluated for normality and heteroscedasticity using Shapiro-Wilk and Breusch-Pagan tests, respectively. Correlation between histological scores and FAP or αSMA staining intensity was determined via linear regression. Gene expression was analyzed using linear mixed-effects models with treatment group (untreated, FAPi, Fasudil), TGFβ stimulation (+/−TGFβ−1), and their interaction as fixed effects and biologic replicate as a random effect. Analysis of residuals and post-hoc comparisons were performed as described above. For cell morphology outcomes, individual cell values were tested for outliers using the ROUT method (Q = 1%). Cell values were then averaged for each donor. Donor averages along with focal adhesion, collagen gel contraction, and nascent matrix data were tested for outliers using Grubb’s method (α = 0.05). Outliers were excluded from analysis. Groups were then analyzed using two-way ANOVAs with Tukey’s procedure for post-hoc testing. All data were checked for normality using Shapiro-Wilk tests. Before post-hoc testing, residuals were checked for normality using Shapiro-Wilk tests and QQ plots. Statistical analyses were performed in R (Version 4.5.2) or GraphPad Prism (Version 10.3.0).

Results

FAP is upregulated in OA synovium across species

Immunofluorescent staining of synovium from small and large animal models of surgically induced OA showed increased FAP protein levels after injury (Fig. 1A–E). Murine and porcine synovium harvested 8 weeks and 6 months after DMM, respectively, demonstrated significantly greater FAP staining intensity compared to synovium from intact control limbs (Murine mean of differences: 10.3 RFU, 95% Confidence Interval: [3.0, 17.6], p = 0.0170; Porcine 8.7 RFU [4.6, 12.8], p = 0.0008) (Fig. 1 H, I). This increase in FAP was also observed in human patients with late-stage OA (Kellgren-Lawrence grade 3 or 4) relative to synovium from healthy controls (24.5 RFU [11.8, 37.3], p = 0.0024) (Fig. 1F, G, J). Notably, the spatial distribution of FAP expression differed across species. In human and porcine samples, FAP staining was primarily localized to the intimal lining, consistent with expansion and activation of lining-layer FLS, characteristic of OA progression. In contrast, murine samples exhibited a more diffuse expression pattern, with FAP-positive cells distributed throughout the synovium, outer meniscus, and subchondral bone.

Fig. 1.

FAP is upregulated in diseased OA synovium across species. (A) Schematic of study design. Representative images of (B) healthy and (C) pathologic murine synovium (n = 5/group) taken from intact and surgically destabilized joints, respectively. Representative images of (D) healthy and (E) pathologic porcine synovium (n = 8/group) taken from intact and surgically destabilized joints, respectively. Representative images of (F) healthy (n = 3) and (G) pathologic (n = 9) human synovium taken from healthy donors and TKA patients, respectively. Synovium is stained for FAP (green) and αSMA (red). Scale bar (merged images): 250μm. Scale bar (individual channels): 50μm. Quantification of FAP and αSMA staining intensity for (H) murine, (I) porcine, and (J) human synovium. MFI = Mean fluorescence intensity. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

FAP expression correlates with mechanoactivation and disease progression

To better understand the time course of FAP expression after joint injury, staining intensity was quantified at 6 weeks and 6 months following destabilization of the medial meniscus in the Yucatan minipig model. These time points were selected as representing early and established phases of disease, based on changes in cartilage mechanical properties [26–28]. FAP staining significantly increased in DMM synovium at both the early (13.7 RFU [6.4, 21.1], p = 0.0013) and late (8.7 RFU [2.5, 15.0], p = 0.0105) time points (Fig. 2B). Samples were also stained for αSMA as a surrogate for mechanoactivation given that myofibroblastic cells express higher levels of this protein. αSMA staining intensity was higher in injured joint synovium at 6 weeks (6.5 RFU [3.6, 9.4], p = 0.0003) and 6 months (7.3 RFU [4.9, 9.8], p < 0.0001) (Fig. 2C). Expression of both proteins was slightly higher at 6 weeks compared to 6 months; however, this difference was only significant for FAP. FAP and αSMA staining intensity positively correlated across individual donors (β = 1.4 RFU/RFU [0.8, 2.1], R² = 0.4, p = 0.0002) (Fig. 2D).

Fig. 2.

FAP expression correlates with mechanoactivation and disease progression in porcine OA. (A) Schematic of study design. Quantification of (B) FAP and (C) αSMA staining intensity. MFI = Mean fluorescence intensity. (D) Correlation between FAP and αSMA staining for individual donors. (E) Representative images of porcine synovium harvested from control and DMM joints 6 weeks (n = 7 control, n = 5 injured) or 6 months (n = 8 control, n = 8 injured) after surgery. Scale bar: 200μm. (F) Correlation between FAP staining and histological scores for inflammatory infiltration (0–3). (G) Correlation between FAP staining and histological scores for hyperplasia (0–3). (H) Correlation between αSMA staining and inflammatory infiltration (0–3). (I) Correlation between αSMA staining and hyperplasia (0–3). *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

To evaluate the relationship between FAP expression and synovial pathology, additional sections were stained with H&E and scored for intimal hyperplasia, vascularity, inflammatory infiltration, and fibrosis. Consistent with prior studies [26–28], DMM resulted in marked synovitis and fibrosis at both 6 weeks and 6 months (Fig. 2E). Regression analysis relating staining intensity with histological scores showed that FAP expression positively correlated with inflammatory infiltration (22.7 RFU/Inflammation Score [9.5, 35.9], R² = 0.3, p = 0.0015) and intimal hyperplasia (7.5 RFU/Hyperplasia Score [1.8, 13.2], R² = 0.2, p = 0.0123) (Fig. 2F, G). αSMA staining intensity positively correlated with inflammatory infiltration (9.6 RFU/Inflammation Score [3.4, 15.9], R² = 0.2775, p = 0.004), intimal hyperplasia (4.0 RFU/Hyperplasia Score [1.5, 6.5], R² = 0.3, p = 0.0026), and vascularity (5.8 RFU/Vascularity Score [1.6, 10.1], R² = 0.2, p = 0.0093) (Fig. 2H, I, Fig. S7). No significant correlations were observed between FAP staining intensity and vascularity or fibrosis (Fig. S7).

Culture on stiff substrates with TGFβ activates FLS in vitro

To establish an in vitro model of FLS mechanoactivation, FLS were seeded on 15kPa polyacrylamide (PA) gels (soft substrate) or glass (stiff substrate). Within each substrate group, half of the samples were additionally stimulated with TGFβ−1 to induce mechanoactivation and promote the transition to a myofibroblastic phenotype [29] (Fig. 3A). FLS on glass showed a significant increase in cell area relative to cells on 15kPa PA gels (1127μm² [844.8, 1409], p < 0.0001). Cell area also trended higher with TGFβ treatment on both substrates (277.2μm² [5.2, 559.5], p = 0.0535) (Fig. 3B, D). In addition to changes in cell size and shape, there were also notable differences in focal adhesion morphology. Cells plated on glass had more distinct focal adhesions and TGFβ supplementation further increased their number and size (Fig. 3C). Cell solidity, a measure of compactness, decreased significantly on glass, but was unchanged by TGFβ treatment. The aspect ratio of FLS did not change with substrate stiffness or TGFβ treatment (Fig. S8).

Fig. 3.

Culture on stiff substrates in the presence of TGFβ induces a mechanically activated phenotype in FLS. (A) Schematic of study design. (B) Representative images of bovine synovial cells cultured on 15kPa polyacrylamide (PA) gels or glass with or without TGFβ (n = 5/group). Scale bar: 50μm. (C) Representative images of focal adhesion staining from the same groups. Scale bar: 50μm. Quantification of (D) cell spread area, (E) actin:αSMA colocalization, and (F) FAP staining intensity. MFI = Mean fluorescence intensity. RT-qPCR showing deltadeltaCT values for (G) αSMA, (H) CTGF, and (I) FAP (relative to soft-cultured non-TGFβ stimulated controls). *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

When fibroblasts become activated, αSMA is incorporated into actin filaments to generate contractile stress fibers [30]. The spatial colocalization (measured as a correlation coefficient) of actin and αSMA staining can therefore be used as a measure of myofibroblastic differentiation, where higher values indicate more incorporation of αSMA into stress fibers [15]. Actin:αSMA colocalization increased significantly with TGFβ treatment (0.2 [0.005, 0.5], p = 0.0474) on both soft and stiff substrates, as did FAP staining intensity (0.02 RFU [0.008, 0.03], p = 0.0076) (Fig. 3E, F). At the transcriptional level, αSMA expression was significantly higher on glass both in the presence (2.7 ΔΔCt [1.3, 4.1], p = 0.0013) and absence (5.1 ΔΔCt [3.6, 6.5], p < 0.0001) of TGFβ. αSMA expression also increased in response to treatment with TGFβ on both soft (4.3 ΔΔCt [2.8, 5.7], p < 0.0001) and stiff (1.9 ΔΔCt [0.5, 3.3], p = 0.0118) substrates (Fig. 3G). Connective tissue growth factor (CTGF, a downstream readout of cell mechanoactivation) demonstrated the same changes in expression (Fig. 3H). PRG4 expression trended downward on glass (−1.9 ΔΔCt [−3.8, 0.03], p = 0.0527), but was restored by TGFβ stimulation, consistent with our previous findings [15] (Fig. S9). Interestingly, both FAP and fibronectin (FN) expression also increased with TGFβ, but were lower on glass compared to softer substrates (Fig. 3I, Fig. S9).

FAP inhibition rescues the morphology of activated FLS

As culture on a stiff substrate with TGFβ−1 resulted in morphological and transcriptional evidence of mechanoactivation (including larger and more numerous stress fibers, greater actin:αSMA correlation, and higher fibrotic gene expression), we selected this culture condition to evaluate the effects of FAP inhibition. Treatment with the FAP inhibitor Ac-Gly-BoroPro (FAPi) was compared to Fasudil, given our previous data showing that this ROCK inhibitor could block mechanoactivation and restore aspects of FLS homeostatic function [15] (Fig. 4A). In all three groups – control, FAPi-treated, and Fasudil-treated – TGFβ significantly increased cell area (1015μm² [84.4, 1946], p = 0.0269), αSMA staining intensity (0.06 RFU [0.02, 0.1], p = 0.0042), and actin:αSMA colocalization (0.4 [0.2, 0.5], p < 0.0001). FAPi treatment significantly reduced αSMA intensity (−0.07 RFU [−0.1, - 0.01], p = 0.012) and actin:αSMA colocalization (−0.2 [−0.4, −0.05], p = 0.0046) but did not impact cell area (Fig. 4B–E). Fasudil treatment also reduced αSMA staining intensity and actin:αSMA colocalization, but the differences did not reach statistical significance. Cell aspect ratio decreased with TGFβ in all three groups, but was not affected by FAPi or Fasudil treatment. No differences in FAP staining intensity or cell solidity were observed across groups (Fig. S10). Transcriptionally, FAPi significantly reduced αSMA expression both in the presence (−3.3 ΔΔCt [−5.0, −1.6], p = 0.004) and absence of TGFβ (−2.9 ΔΔCt [−4.9, −0.8], p = 0.0057) (Fig. 4F). CTGF expression increased with TGFβ stimulation (4.6 ΔΔCt [2.8, 6.5], p < 0.0001), and decreased slightly with FAPi, though this did not reach statistical significance (Fig. 4G). FAP and FN expression levels followed a similar pattern (Fig. 4H, S11). PRG4 expression was unaffected by treatment with FAPi or Fasudil (Fig. S11).

Fig. 4.

Inhibition of FAP partially rescues the morphological and transcriptional profile of activated FLS. (A) Schematic of study design. FUNCAT = functional noncanonical amino acid tagging. (B) Representative images of bovine synovial cells left untreated (n = 5/group) or treated with FAPi (n = 5/group) or Fasudil (n = 4/group) ± TGFβ. Scale bar: 50μm. Quantification of (C) cell spread area, (D) αSMA staining intensity, and (E) actin:αSMA colocalization. MFI = Mean fluorescence intensity. RT-qPCR showing deltadeltaCT values for (F) αSMA, (G) CTGF, and (H) FAP (relative to non-TGFβ stimulated controls). *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

Targeting FAP impedes focal adhesion formation

Mature focal adhesions are essential for myofibroblastic differentiation [31]. Based on reports that FAP colocalizes with integrin α₃β₁ [22,24], we specifically examined the effect of FAPi treatment on focal adhesion morphology. FAP inhibition significantly reduced the number of focal adhesions per cell (−374.2 FA/Cell [−534.4, −213.9], p < 0.0001), total focal adhesion area per cell (−909.7μm² [−1139, −680.7], p < 0.0001), average focal adhesion length (−0.4μm [−0.8, −0.08], p = 0.0119), and average focal adhesion area (−0.6μm² [−0.9, −0.2], p = 0.0008), both with and without TGFβ (Fig. 5A–C, Fig. S12). Notably, TGFβ stimulated cells that were treated with FAPi had fewer and smaller focal adhesions than non-stimulated control cells. Treatment with FAPi also reduced the percentage of focal adhesions greater than 8μm (−0.7% [−1.3, −0.05], p = 0.032) – a length indicative of a “supermature” focal adhesion [31] (Fig. 5D). Fasudil similarly decreased focal adhesion number, length, and area (Fig. 5A–D, Fig. S12).

Fig. 5.

FAP inhibition reduces focal adhesion formation and maturation. (A) Representative images of focal adhesions in bovine synovial cells left untreated (n = 4/group) or treated with FAPi (n = 4/group) or Fasudil (n = 3/group) ± TGFβ. Scale bar: 50μm. Quantification of (B) focal adhesions per cell, (C) total focal adhesion area per cell, and (D) percentage focal adhesions greater than or equal to 8μm. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

FAP inhibition reduces contractility and fibrogenic capacity of FLS

After interrogating the morphological and transcriptional effects of FAP inhibition, we next sought to investigate whether FAPi could reverse the functional consequences of mechanoactivation. To evaluate FLS contractility, cells were seeded in Type I collagen gels and percent contraction was quantified over time. As expected, gels stimulated with TGFβ contracted faster than non-stimulated gels. At day 4, both stimulated and non-stimulated samples treated with FAPi contracted significantly less than controls (20% Area Change [30, 0.2], p = 0.0461) (Fig. 6A, B). Fasudil-treated samples also showed less contraction at day 4 (30% Area Change [40, 10], p = 0.0002).

Fig. 6.

FAP inhibition reduces contractility and fibrogenic capacity of activated FLS. (A) Representative images of collagen gels seeded with bovine synovial cells and left untreated (n = 7/group) or treated with FAPi (n = 7/group) or Fasudil (n = 6/group) ± TGFβ. Scale bar: 1 mm. (B) Quantification of collagen gel contraction. (C) Representative images of nascent matrix labeling. Scale bar: 250μm. (D) Quantification of nascent matrix produced by untreated bovine synovial cells (n = 6/group) or those treated with FAPi (n = 6/group) or Fasudil (n = 6/group) and cultured in monolayer for 6 days. *p < 0.05. ***p < 0.001.

To evaluate the fibrogenic capacity of FLS, we measured nascent matrix production using functional noncanonical amino acid tagging (FUNCAT) [32]. Treatment with TGFβ increased nascent matrix production in control, FAPi-treated, and Fasudil-treated samples (152586 RFU [1348,303824], p = 0.047). While both Fasudil and FAPi treatment appeared to reduce matrix production, this did not reach the level of significance (Fig. 6C, D).

Discussion

Despite the significant global burden, there are currently no disease-modifying therapeutics for OA [33]. With the recognition of OA as a joint-wide disease, synovial pathology has been identified as a potential driver of joint degeneration [7,15,34,35]. As OA progresses, the synovium undergoes marked structural remodeling and loses key homeostatic functions [3,36]. Fibrotic changes, mediated by mechanoactivated FLS, are thought to drive further fibrosis, inflammation, and catabolism throughout the joint [35]. The protease FAP is a marker of inflammation and fibrosis that is upregulated in human OA synovium [37]. In addition to its proteolytic functions, we hypothesized that FAP contributes to OA progression by promoting FLS mechanoactivation. Thus, the purpose of this study was to confirm the upregulation of FAP in OA synovium across species, evaluate the regulation of FAP expression by FLS when challenged by activating cues, and determine whether FAP inhibition could rescue the homeostatic phenotype of activated FLS.

Here, we show that the increase in FAP seen in human OA synovium [37] is replicated in both small and large animal models. The correlation between FAP and αSMA staining in porcine synovium suggests that cells in more fibrotic, mechanoactivated tissue express higher levels of FAP. FAP staining also positively correlated with intimal hyperplasia and inflammatory infiltration, two histological markers of synovitis, further linking FAP expression and synovial pathology. Localization of FAP differed across species, with expression largely restricted to the intimal lining in human and porcine samples but distributed diffusely in murine samples, likely reflecting species-specific anatomy, disease kinetics, and relative contributions of periarticular tissues to disease progression. Although this variability should be considered when extrapolating mechanistic insights across models, the consistent induction of FAP supports its relevance as a marker of synovial pathology in OA.

Of note, a limitation of this study is that site-specific synovial samples were not obtained from human patients or animal models. Given the known regional heterogeneity of synovial pathology within the joint, it is possible that FAP expression and FLS activation vary by anatomical location and local mechanical environment. Future studies incorporating spatially resolved sampling across defined joint regions will be important in determining whether FAP-mediated mechanoactivation is uniformly present or enriched in specific areas.

After confirming the upregulation of FAP in both our murine and porcine OA models, we turned to an in vitro model of FLS mechanoactivation to understand the regulation of FAP expression. Bovine FLS were activated using both mechanical and chemical stimuli – substrate stiffness and TGFβ−1, respectively. As expected, both stimuli induced cell spreading, actin-αSMA colocalization, and fibrotic gene expression, demonstrating the efficacy of the system in promoting mechanoactivation. We also found that TGFβ−1, but not substrate stiffness, induced FAP expression. The transcription factors EGR-1 and SMAD3, both downstream of TGFβ−1 signaling, have been shown to induce FAP expression via direct interaction with the FAP promoter in human sarcoma and melanoma cells [38,39]; it is likely FAP expression in FLS is controlled by similar mechanisms.

Next, we evaluated the effect of FAP inhibition in FLS cultured on glass, as these cells were in the most mechanoactivated state, as defined by morphological and transcriptional outcomes, and represent the extreme pathological phenotype. Using this system, we found that FAPi treatment reduced αSMA production and the incorporation of αSMA into contractile stress fibers, ultimately decreasing FLS contractility when seeded in Type I collagen gels, even in the presence of TGFβ−1. These findings support a paradigm in which FAP functions not only as a degradative enzyme contributing to cartilage breakdown, but also as a TGFβ-responsive effector that promotes an activated and fibrotic FLS phenotype. Given that synovial pathology is an early and potentially modifiable feature of OA, targeting FAP may offer a dual benefit—reducing cartilage degradation and restoring synovial homeostasis. Importantly, FAP inhibition appeared to be nearly as effective as Rho/ROCK inhibition in reversing features of FLS mechanoactivation, but may represent a more targeted approach with fewer systemic side effects, given that FAP is generally not expressed under physiologic conditions [40].

While it is clear that FAP is associated with the mechanoactivated synovial cell state, the precise mechanism by which it mediates this process is not fully determined. We observed a marked reduction in focal adhesion number, length, and area with FAP inhibition. Focal adhesions play a vital role in mechanotransduction by linking the actin cytoskeleton to the ECM [41]. Using a fibrosarcoma cell line, Baird et al. [19] recently showed that the role of FAP in the induction of adhesion, migration, and invasion is mediated by β₁ integrins, one of the cell surface proteins that make up the ECM-binding domain of focal adhesions [30]. Baird et al. [19] also showed that FAP expression promotes activation of focal adhesion kinase (FAK). Upon binding to integrins, FAK directs focal adhesion maturation and the cytoskeletal remodeling required to transmit force from the ECM to the nucleus [42]. Given our data showing a decrease in focal adhesion number and maturity with FAP inhibition, it is possible that FAP promotes mechanoactivation of FLS via a similar mechanism, upregulating FAK signaling through interaction with integrins.

Interestingly, despite being a marker of fibrosis, FAP gene expression was lower on glass than on soft PA gels. This suggests that while stiffness promotes myofibroblastic differentiation, FAP induction is not a direct result of stiffness-driven mechano-sensing. Instead, our data suggest that FAP expression may be more closely tied to biochemical signals such as TGFβ−1 or matrix composition. Consistent with this, others have reported decreased FAP expression with prolonged culture on stiff tissue-culture plastic [18], and Avery et al. [43] found that soft, fibronectin-rich ECMs promoted a FAP-high, αSMA -low phenotype, while stiff, collagen I-rich ECMs promoted a FAP-low, αSMA -high phenotype in murine lung fibroblasts. While additional studies are needed to further define the distinct contributions of matrix stiffness and composition to FAP regulation, these findings support a model in which FAP expression is driven primarily by profibrotic biochemical cues and integrates with mechano-signaling pathways to promote FLS activation and synovial pathology (Fig. 7).

Fig. 7.

Proposed role of FAP in myofibroblastic differentiation. In healthy joints, quiescent FLS reside within a soft, compliant ECM, exhibit minimal cytoskeletal tension, and maintain homeostatic functions such as the production of lubricin (PRG4). During progression of OA, synovial fibrosis leads to ECM stiffening, which is sensed by FLS through integrin engagement and focal adhesion assembly. Simultaneously, increased bioavailability of the profibrotic cytokine TGFβ induces upregulation of FAP. Our data suggest that FAP supports focal adhesion formation and/or maturation, enhancing mechanotransduction, amplifying cytoskeletal contractility and promoting myofibroblastic differentiation. YAP = Yes-associated protein. TAZ = Transcriptional coactivator with PDZ-binding motif. 14–3–3 = 14–3–3 adaptor protein. ROCK = Rho-associated protein kinase. TEAD = Transcriptional enhanced associate domain. CCN2 = Connective tissue growth factor. CYR61 = Cysteine-rich angiogenic inducer 61. ACTA2 = α-smooth muscle actin.

It also remains unclear whether FAP mediates FLS mechanoactivation via its enzymatic or non-enzymatic functions. FAP cleaves a wide range of substrates including collagen I, collagen II, FGF-21, and a2-antiplasmin [44,45], and soluble FAP has been shown to promote cartilage degradation in OA via cleavage on MMP-degraded collagen II [17]; however, several studies have also identified non-proteolytic actions of FAP. In bone marrow mesenchymal stromal cells (BM-MSCs) FAP depletion inhibited migration, but treatment with 2 distinct protease inhibitors failed to reproduce this loss of migratory capacity [20]. Here we used Ac-Gly-BoroPro, a competitive inhibitor which binds directly to the catalytic active site of FAP [46]. Although this suggests the enzymatic function of FAP is required to promote FLS mechanoactivation, there is also the potential for competitive inhibitors to induce conformational changes that interfere with the non-proteolytic functions of FAP. Further studies are needed to evaluate the effects of different FAP inhibitors on both the enzymatic and non-enzymatic functions of this protein and to elucidate the precise mechanism by which FAP mediates mechanoactivation.

In summary, our study identifies FAP as a conserved, TGFβ-responsive mediator of FLS activation and fibrosis in OA that is uniquely present in disease. FAP inhibition partially restored a homeostatic FLS morphology, contractility, and matrix production, and so targeting FAP may offer a novel strategy to modify disease progression and improve overall joint health in OA. As the tissue crosstalk and mechanical forces present in the joint are impossible to recapitulate in vitro, it will also be critical to assess outcomes of FAP inhibition in vivo. Future in vivo studies will evaluate the effect of FAP inhibition on synovial fibroblast phenotype, synovial stiffness and structure, cartilage integrity, and pain behavior, as well as potential synergistic effects of targeting both mechanobiological and inflammatory pathways in a large animal OA setting.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.joca.2026.03.121.

Data Availability

Data available upon request.